BIOLOGICAL OVERVIEW

Connectin, a homophilic cell adhesion protein expressed in a subset of muscles and the motoneurons that innervate them, was isolated and characterized by two different laboratories at the same time using two completely different methods. One investigation has been seeking targets for homeotic genes. An immunopurification approach was used that enhances the formation of short chromatin fragments through their association with endogenous Ultrabithorax protein. Embryonic nuclei were digested with a restriction enzyme and then lysed: the soluble chromatin was affinity-purified against a matrix containing antibodies to UBX proteins, and the DNA fragments from the immunopurified chromatin were then cloned (Gould, 1992). A second investigation employed brute force techniques: 11,000 enhancer trap lines (lines in which P elements, bearing the beta-galactosidase gene, have been inserted at random in the genome) were screened for beta-galactosidase expression in subsets of muscle fibers prior to innervation. In such lines, beta-galactosidase expression is subject to regulation by the different promoters into which the P element inserts. Two of these inserts were in Connectin and Toll, both members of the leucine-rich repeat family (Nose, 1992).

Drosophila Connectin (Con) is a cell surface protein of the leucine-rich repeat family. Other Drosophila proteins sharing the LRR repeat include Slit and Chaoptin. During the formation of neuromuscular connectivity, Con is expressed on the surface of a subset of embryonic muscles and on the growth cones and axons of the motoneurons that innervate these muscles, including primarily segmental nerve a (SNa) motoneurons and their synaptic targets (lateral muscles). In vitro, Con has been shown to mediate homophilic cell adhesion (Nose, 1992).

Transgenic lines have been generated that ectopically express Con on all muscles. Ectopic Con expression does not result in gross developmental defects of the CNS, PNS and musculature. Major motor nerves project normally in the CNS and in the periphery. All muscles form in their correct locations with normal insertion sites (Nose, 1997). This result is rather extraordinary considering the complex pattern of Con expression on neurons and the muscles they innervate. From stages 12 through 16 in development Con is expressed on the surface specific ventral and lateral muscles and on the segmental and intersegmental nerves that innervate them. Con is also expressed on the surface of glial cells (glial cells are thought to provide directional clues for axonal extention) (Nose, 1992).

In the transformant embryos and larvae, where Con is ectopically expressed in all muscles, close examination reveals that SNa motoneurons often inappropriately innervate a neighboring non-target muscle (muscle 12) that ectopically expresses Con. It was first thought that inappropriate Con expression indicated a repulsive function. Motoneurons change both their morphology and their trajectory when they encounter ectopic Con-positive ventral muscles, displaying 'bypass," "detour," and "stall" phenotypes (Nose, 1994). However later studies in which Fas II and Fas III were inappropriately expressed revealed a similar phenotype: abnormalities in the innervation of specific muscles. It was reasoned that since all three proteins (Con, Fas II and Fas III) give similar phenotypes then these phenotypes must be the result of an indirect influence of increased muscle adhesion arising from the mis-expression of the proteins and not the result of repulsive effects. Furthermore, the ectopic synapse formation was shown to be dependent on the endogenous Con expression on the SNa motoneurons. These results show convincingly that Con can function as an attractive and homophilic target recognition molecule in vivo (Nose, 1997).

beaten path (beat) functions at specific choice points along the major motor nerves where subsets of motor axons defasciculate and then steer into their muscle target regions. In beat mutant embryos, motor axons fail to defasciculate and consequently bypass their targets. This phenotype is suppressed by mutations in FasII and Con, suggesting that beat provides an antiadhesive function, possibly to counteract the adhesive functions of FasII and Connectin. beat encodes a novel secreted protein that is expressed by motoneurons during outgrowth. It is suggested that Beat protein is secreted by motor axons where it functions to regulate their selective defasciculation at specific choice points (Fambrough, 1996).

Hox function is required for the development and maintenance of the Drosophila feeding motor unit

Feeding is an evolutionarily conserved and integral behavior that depends on the rhythmic activity of feeding muscles stimulated by specific motoneurons. However, critical molecular determinants underlying the development of the neuromuscular feeding unit are largely unknown. This study identified the Hox transcription factor Deformed (Dfd) as essential for feeding unit formation, from initial specification to the establishment of active synapses, by controlling stage-specific sets of target genes. Importantly, Dfd was found to control the expression of functional components of synapses, such as Ankyrin2-XL, a protein known to be critical for synaptic stability and connectivity. Furthermore, Dfd was uncovered as a potential regulator of synaptic specificity, as it represses expression of the synaptic cell adhesion molecule Connectin (Con). These results demonstrate that Dfd is critical for the establishment and maintenance of the neuromuscular unit required for feeding behavior, which might be shared by other group 4 Hox genes (Friedrich, 2016).

Stereotypical motor behaviors are the primary means by which animals interact with their environment, forming the final output of most CNS activity. One such behavior is feeding, a crucial and highly conserved activity in all animals. The motor output consists of coordinated contractions of distinct head muscles in a rhythmic pattern required for chewing, sucking, and swallowing of food. Food uptake in adult flies has recently been shown to be controlled by a single pair of interneurons emanating from the subesophageal ganglion (SEG), an insect brain region primarily associated with taste and feeding. Although a substantial number of neurons are linked to different aspects of feeding behavior in flies, molecular factors critical for the establishment and development of feeding motor patterns have not been identified so far (Friedrich, 2016).

The fruit fly Drosophila melanogaster is an excellent model to study the developmental aspect of feeding behavior for several reasons. First, Drosophila takes up food extensively during its larval stage, when the organism almost exclusively feeds to increase its body weight and size. Additionally, the anatomical framework and motor patterns critical for larval food uptake are well described. Feeding requires the rhythmic extension and retraction of the head skeleton, the cephalopharyngeal skeleton (CPS), coupled with coordinated elevation and depression of the mouth hooks (MHs), mandible-derived structures required for chopping up solid food, and subsequent food ingestion. The repetitive larval feeding movements are controlled by head muscles innervated by CNS nerves emerging from the SEG. CPS protraction and tilting are mediated by protractor muscles receiving input from the prothoracic nerve, while MH motor patterns are controlled by the mouth hook elevator (MHE) and depressor (MHD), which are innervated by the maxillary nerve. Food ingestion is achieved by the cibarial dilator muscle (CDM), which is connected to the CNS via the antennal nerve. The cellular framework of Drosophila larval feeding is established during embryogenesis. Thus, molecular and genetic approaches can be used to identify and analyze factors controlling specification and communication of cell types critical for larval motor patterns. In contrast, neuromuscular units required for motor activities in adult flies develop from stem cell systems during larval and pupal stages in a process called metamorphosis. Due to the limited accessibility of this transitional phase, embryonic stages are better suited to study the development of neuromuscular units required for regional movements (Friedrich, 2016).

The Hox family of transcription factors (TFs) have emerged as key regulators of motor behaviors. One such behavior is locomotion, which Drosophila larvae perform by region-specific contractions of abdominal segments allowing them to crawl on substrate. Segment-specific changes of peristaltic movements in animals carrying mutations in the Hox genes Ultrabithorax (Ubx) and abdominal-A (abd-A) led to the assumption that Hox genes orchestrate the development of regional motor activities. Recent studies have now revealed that Hox proteins perform their task in a very refined manner and seem to have a direct transcriptional input on successive steps of motoneuronal development. As one such example, Hox5 function was shown to be required in motoneurons that control the contraction of breathing muscles in vertebrates: Hox5 deletion in mice leads to progressive death of phrenic motor column (PMC) neurons as well as to the inability of surviving PMC neurons to innervate the diaphragm muscle. However, despite the fact that blocking motoneuron apoptosis did rescue the decline in PMC neuron number, branching and innervation defects were still unchanged under these conditions. These findings imply that Hox5 proteins directly regulate early and late processes in the course of PMC neuron differentiation, a hypothesis still awaiting confirmation (Friedrich, 2016).

Hox genes are segmentally expressed along the anterior-posterior body axis of animals, suggesting that members of this gene family expressed in the head region should control food uptake. Intriguingly, the Drosophila group 4 Hox gene Deformed (Dfd), which is known to specify the SEG, has already been associated with feeding behavior before: animals carrying a hypomorphic Dfd allele starve to death as adult flies due to the inability to move their proboscis, an action crucial for food ingestion. This work reveals that Dfd, which is expressed in many cell types, including a large number of SEG neurons, is functional in motoneurons and muscles that drive the movements critical for hatching and feeding. Most interestingly, it was shown that Dfd exerts its function via direct and stage-specific regulation of target genes. Importantly, Ank2-XL, a microtubule organizing protein required for synaptic stability, was shown to be under direct Dfd control throughout different stages in the animal's life. Furthermore, synchronous expression of Dfd targets with critical function in synaptic target specificity, in particular, the cell adhesion molecule (CAM) Connectin (Con), was found in feeding neurons and muscles. This suggests that Dfd positively and/or negatively regulates different CAMs providing a specificity code required for the establishment of regional motor units (Friedrich, 2016).

Consistent with a temporal requirement of Dfd for motoneuronal development, an over-representation of neuronal genes was found among those genes associated with ChIP-seq identified Dfd binding regions, which were classified as Dfd target genes. Importantly, grouping of these genes based on similar GO annotations showed that they operate at different time points in neuronal development: during neurogenesis and neuronal specification (32/182), when axon outgrowth and guidance decisions occur (86/182), and during synapse-related processes (85/182). To test the temporal control of these genes by Dfd, their expression was analyzed when Dfd function was abolished at two different developmental stages. For early interference, ^Dfd16 loss-of-function embryos was used, while late interference was achieved using animals that carry the temperature-sensitive ^Dfd3 allele and were shifted to the restrictive temperature only during larval stages. Early neurogenesis target genes, including prospero, a gene involved in asymmetric neuroblast division, was found to be mis-localized in ^Dfd16 null mutant embryos. Consequently the expression of genes required for subsequent processes in motoneuronal development, like the axon guidance genes capricious (caps), roundabout 2 (robo2), roundabout 3 (robo3), and Neural Lazarillo (NLaz), were also affected. Thus, ^Dfd16 null mutants are unable to form the neuromuscular unit required for MH movements due to the inability to activate the proper developmental program. In contrast, the MH-associated motor unit of ^Dfd3 animals shifted to the restrictive temperature during larval stages was intact, with respect to outgrowth of maxillary nerve projecting motoneurons and MHE innervation. Accordingly, the expression of early Dfd neuronal targets was unchanged (data not shown). However, compared to the control group the expression of Dfd target genes critical for synapse-related processes was substantially altered in these late-shifted ^Dfd3 third-instar larvae. This includes Ankyrin2 extra large (Ank2-XL), which is encoded in the ank2 locus. Ank2-XL, which is part of a membrane-associated microtubule-organizing complex, is known to be required for the establishment of appropriate synaptic dimensions and release properties (Stephan, 2015). This study found that not only was Ank2 mRNA levels reduced in SEG neurons in late-shifted ^Dfd3 third-instar larvae, but also observed decreased Ank2-XL protein expression in synaptic boutons, axons and their terminals on the MHE. Similar to a recent report (Stephan, 2015), Futsch/MAP1B, a microtubule-associated protein known to form a membrane-associated complex with Ank2-XL, was also found to be reduced in synaptic boutons of late-shifted ^Dfd3 third-instar larvae. Concomitantly, the morphology of synaptic boutons on this muscle was also changed in late-shifted ^Dfd3 third-instar larvae: they were not of uniform size but appeared often dramatically increased compared to boutons of control animals. This is in line with the described phenotype of ank2-XL mutant animals (Stephan, 2015), which was suggested to reflect the failed separation of neighboring boutons. The effects observed are due to Dfd's action in (moto)neurons, as tissue-specific knockdown of Dfd activity in neuronal cells only using the elav-GAL4 driver in combination with two independent UAS-DfdRNAi lines resulted in severe bouton phenotypes and Ank2-XL expression changes, while the muscle architecture was completely normal. Similar results on Ank2-XL expression and synapse morphology were obtained in Dfd13/Df(3R)Scr third-instar larvae that survived to this stage. The effect of Dfd on synapses on the MHE is specific, since neuromuscular junctions on control muscles, like the CDM, were completely normal with respect to their morphology and Ank2-XL expression in late-shifted ^Dfd3 third-instar animals. These results show that Dfd activity is continuously required during the formation of the feeding motor unit, from its specification to the establishment of synaptic connections, and that Dfd executes this function by directly regulating the transcription of phase-specific components. Intriguingly, these findings demonstrate that the Hox TF Dfd is one of the upstream regulators coordinating Ankyrin-dependent microtubule organization and synapse stability and provides evidence that Dfd function is required even after the initial establishment of the motor unit to control synapse-related processes via its synaptic targets, like Ank2-XL. Finally, the results indicate that synaptic stability and plasticity is not only determined by the half-life of synaptic proteins, but is dependent on a robust transcriptional program that provides a continuous supply of essential synaptic components that maintain the system (Friedrich, 2016).

This analysis has shown that Dfd is expressed in SEG motoneurons. In addition, Dfd was found to be present in embryonic muscles, which later form the feeding/hatching motor unit. Defects were observed in the structure and number of the MH-associated muscles in the embryo when Dfd function was abolished. As was the case in the CNS, Dfd seems to execute its muscle-specific function in an immediate manner, since a substantial fraction (7.4%) of the genome-wide identified Dfd target genes is associated with mesoderm-related functions. The innervation of the Dfd-expressing MHE by Dfd-positive motoneurons raised the intriguing possibility that the activity of the Hox protein Dfd provides a code on the functionally connected neurons and muscles crucial for the recognition and matching of the synaptic partners and, thus, the execution of rhythmic motor patterns. Consistent with this hypothesis, 27 of the ChiP-seq identified Dfd target genes encode factors with described functions in muscles and the nervous system, and importantly nine of these genes play an important role in synaptic target recognition, like tartan (trn), Connectin (Con), or capricious (caps). Therefore, the expression of the homophilic cell adhesion molecule Con was analyzed, and it was found to be exclusively expressed in motoneurons and muscles devoid of Dfd protein in wild-type embryos, suggesting that Dfd might function as a suppressor of Con expression. In order to provide vigorous proof for this hypothesis, cells were specifically labeled that were devoid of Dfd function in Dfd mutants, and their ability to now express Con was analyzed. Use was made of the fact that ^Dfd16 mutants that do not produce any functional protein (protein-null mutants) still express Dfd mRNA. Consistent with this hypothesis, de-repression of Con mRNA expression was found in many Dfd mutant neuronal cells that were labeled by the presence of Dfd mRNA. Due to the inability of Dfd mutant embryos to involute their heads (which reorganizes the order of the head muscles) and due to the high variance of the muscle phenotype, Con expression in this tissue was not shown in the Dfd loss-of-function situation. Taken together, these results demonstrate that Dfd is one of the critical upstream regulators, which coordinates the interdependent events of neuromuscular development and connectivity by positively or negatively regulating the expression of synaptic target selection molecules on the interacting motoneurons and muscles. Furthermore, it shows that the expression of synaptic cues is tightly regulated even in neurons located in close or direct proximity, allowing these cells to express different sets of synaptic recognition molecules thereby ensuring that they make the proper connections with their synaptic partners (Friedrich, 2016).

Hox genes have been shown to control several motor activities along the anterior-posterior axis of animals; however, critical determinants regulating feeding movements have not been previously identified. This study has shown that the Drosophila group 4 Hox gene Dfd controls multiple aspects in both the establishment and maintenance of the neural network controlling feeding behavior (Friedrich, 2016).

A crucial finding from this study is that Hox TFs are required throughout the formation of regional motor units and mediate their effect not only through the induction of downstream TFs. In fact, it was shown that Hox factors control distinct effector target genes, which realize stage-specific processes in a very immediate manner. This is true for Ankyrin2-XL, which, along with the MAP1B homolog Futsch, forms a membrane-associated microtubule-organizing complex that determines axonal diameter, supports axonal transport, and controls synaptic dimensions and stability (Stephan, 2015). Interestingly, it was found that Dfd is required for the maintenance of Ank2-XL expression, not only when the motor system is established but also when it is fully operational. In the light of recent findings showing that mis-regulation of Ankyrin 1 (ANK1) has an important role in the neurodegenerative Alzheimer disease, these results raise the intriguing possibility that Hox genes have a neuro-protective function (Friedrich, 2016).

An important question arising from this study is whether the establishment of feeding-related motor patterns is one of the basic functions of group 4 Hox genes and thus conserved in the animal kingdom. Promisingly, it is known that tongue muscles critical for rhythmic feeding movements in mammals are innervated by the hypoglossal nerve. This nerve has its origin in rhombomere 8, which expresses several group 4 Hox genes, including Hoxb4. Preliminary analysis using a previously identified fish Hoxb4 promoter as a reporter in the teleost fish medaka (Oryzias latipes) shows that GFP is expressed in distinct neuronal subpopulations of the post-otic hindbrain and the spinal cord in stable Hoxb4-GFP medaka embryos. Intriguingly, a subset of neurons co-expressing GFP and Hoxb4 project their axons ventrally toward Hoxb4-positive cells within the pharyngeal region. Both the branchial muscles and the pharyngeal jaw specialized for feeding in teleost fish develop from this area. When medaka embryos have developed into hatchlings, axons emerging from GFP-labeled neurons innervate branchial muscles and the sternohyodeus, muscle groups required for mouth opening and food swallowing. Thus, the regulatory and transcriptional network dictating the formation of the respective feeding units in flies and fish could be conserved despite the fact that muscles and bones responsible for the execution of feeding movements are of different origin. In future, more functional studies are needed to validate the potentially conserved role of homology group 4 Hox genes in regulating rhythmic feeding movements throughout the animal kingdom (Friedrich, 2016).

GENE STRUCTURE

Bases in 5' UTR - 4160

Exons - 4

Bases in 3' UTR - 756

PROTEIN STRUCTURE

Amino Acids - 682

Structural Domains

The N-terminal region contains a stretch of hydrophobic amino acids, a characteristic feature of a signal sequence, known to facilitate secretion. Another stretch of approximately 17 amino acids (mostly hydrophobic) at the C-terminal end is characteristic of proteins that are attached to the membrane via a phosphatidylinositol anchor. These amino acids are preceded by others that fulfill the consensus for the cleavage and attachment site for a PI-lipid anchor. Since the protein is expressed on the cell surface and promotes cell aggregation, it is suggested that the protein is attached to the membrane, presumably via a PI anchor. Incubating membranes from Connectin-expressing cultured cells with PI-phospholipase C leads to the release of over 50% of the membrane-bound Connectin (Nose, 1992 and Gould, 1992).

Connectin contains ten stretches of 24 amino acid leucine rich repeats (LRRs). LRRs have been identified in a variety of different proteins from a wide range of species, including human leucine-rich a2-glycoprotein and a noncatalytic domain of yeast adenylate cyclase. These repeats are on average 24 amino acids in length and are characterized by a periodic distribution of hydrophobic amino acids, especially leucine residues, separated by more hydrophilic amino acids. Each repeat unit can potentially adopt an amphipathic structure. Several possible functions have been suggested for LRRs. They could play a role in protein-protein interactions or in mediating interactions between the protein and cellular membranes (Nose, 1992 and Gould, 1992).

In Drosophila, LRRs have been found in Toll, Chaoptin and Slit. Some of the proteins with LRRs are known to share additional amino acid similarity extending to either N-terminal LRR-flanking sequences or C-terminal LRR-flanking sequences, or both. This sequence similarity is found in Connectin in the C-terminal LRR-flanking region but not in the N-terminal flanking region. This similarity is found in Toll and Slit, and in some of the vertebrate LRR proteins including human platelet glycoprotein 1b and oligodendrocyte-myelin glycoprotein. One major characteristic of this region is the four cysteines that are highly conserved among this group of proteins. Connectin, however, lacks the fourth cysteine. A functional role for this region has been demonstrated in vivo; mutations contained within this region in Toll confer a dominant phenotype (Nose, 1992).

Physical attributes of leucine rich repeat (LRR) proteins

Unusual properties are found for a synthetic LRR peptide derived from the sequence of the Drosophila membrane receptor Toll. In neutral solution the peptide forms a gel revealed by electron microscopy to consist of extended filaments approximately 8 nm in thickness. As the gel forms, the circular dichroism spectrum of the peptide solution changes from one characteristic of random coil to one associated with beta-sheet structures. Molecular modelling suggests that the peptide forms an amphipathic structure with a predominantly apolar and charged surface. Based on these results, models for the gross structure of the peptides filaments and a possible molecular mechanism for cellular adhesion are proposed. The finding that Toll-LRR forms intramolecular beta-sheet structures supports the view that LRRs can participate in protein-protein interactions and homotypic cellular adhesion. It could be that LRRs expressed on the cell surface are initially of disordered structure and that interactions with similarly disordered LRRs on an adjacent cell causes the formation of an extended and stable intermolecular beta structure. Such a mechansim could provide a molecular basis for cellular adhesion mediated by LRRs (Gay, 1991).

The crystal structure is present at 2.5 A resolution of the complex between ribonuclease A and Ribonuclease inhibitor (RI), a protein built entirely of leucine-rich repeats. The unusual non-globular structure of RI, its solvent-exposed parallel beta-sheet and the conformational flexibility of the structure are used in the interaction; they appear to be the principal reasons for the effectiveness of leucine-rich repeats as protein-binding motifs. The structure can serve as a model for the interactions of other proteins containing leucine-rich repeats with their ligands (Kobe, 1995).

The horseshoe-shaped structure of a ribonuclease inhibitor (RI), with a parallel beta sheet lining the inner circumference of the horseshoe and alpha helices flanking its outer circumference, is the only X-ray structure containing these repeats to have been determined. Despite the fact that the lengths and sequences of the RI repeats differ from those of the most commonly occurring LRRs, it was deemed worthwhile to derive a three-dimensional structural framework of these more typical LRR proteins, using the RI structure as a template. Sequence alignments of 569 LRRs from 68 proteins were obtained by a profile search and used in a comparative sequence analysis to distinguish between residues with a probable structural role and those which seemed essential for function. This knowledge, along with the known atomic structure of RI, was used to model the three-dimensional structure of the most common LRR units. These modeled units were then used to build the three-dimensional structure of the extracellular domain of the thyrotropin receptor (TSHR)--a "typical" LRR protein. The modeled TSHR structure adopts a non-globular arrangement, similar to that in RI. The beta regions of this typical LRR protein are the same as in the RI structure, whereas the alpha helices are shorter and the conformations of the alpha beta and beta alpha connections are different. As a result of these differences it was not possible to pack together typical LRR units using repeats such as those found in RI. This mutually exclusive relationship is supported by sequence analysis. The predicted structure of the typical LRRs obtained here can be used to build models for any of the known LRR proteins and the approach used for the prediction could be applied to other proteins containing internal repeats (Kajava, 1995).

date revised:  23 May 97  

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.